Helen DeWitt has a book coming out later this year (hurrah!), and having had her editor suggest she make her use of “afterward(s)” and “backward(s)” consistent, she started trying to find out what actual usage was, and having found little satisfaction, she did what any sensible person would do and wrote Language Log to ask. Mark Liberman responds with a most interesting post, providing this summary and then going on to give some numbers:
• The choice between X-ward and X-wards is subject to variation in all regions and registers.
• In the case of back– and after-, American English uses a higher proportion of –ward forms than British English does (and the proportion of afterward forms in British English is very low).
• In the case of backward(s), in both American and British English, conversation shows a lower proportion of –ward forms (and thus a higher proportion of –wards forms) than academic prose does.
• In the case of afterward(s), American English also shows more –wards forms in conversation than in academic prose, while British English seems to go the opposite way, though there are too few –wards forms overall to be confident about the relationship.
• At least in American English, these patterns seem to be stable over time.
As I say in a comment there, I’ve been wondering about this myself—I’ve been enforcing a style guide that requires –ward on the MSS I edit, but I wasn’t sure what the facts of usage were, including my own. Head over there for details and a good discussion, and while you’re there check out this thread discussing Lameen Souag’s post on Libyan dialect features in Gaddafi Jr’s speech.